UCD-BDLab/BioNeuralNet: v1.3.0

Fuente: Zenodo
Saved in:
Bibliographic Details
Main Authors: Vicente Ramos, Mohamed Abdel-Hafiz, SundousHussein
Format: Recurso digital
Published: Zenodo 2026
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866902134880993280
author Vicente Ramos
Mohamed Abdel-Hafiz
SundousHussein
author_facet Vicente Ramos
Mohamed Abdel-Hafiz
SundousHussein
contents <h2>[1.3.0] - 2026-04-01</h2> <h3>Network Module (<code>bioneuralnet.network</code>)</h3> <ul> <li><strong>New dedicated module</strong>: Network construction and analysis moved from <code>bioneuralnet.utils</code> to <code>bioneuralnet.network</code>.</li> <li><strong>Renamed construction functions</strong>: <code>gen_similarity_graph</code> -> <code>similarity_network</code>, <code>gen_correlation_graph</code> -> <code>correlation_network</code>, <code>gen_threshold_graph</code> -> <code>threshold_network</code>, <code>gen_gaussian_knn_graph</code> -> <code>gaussian_knn_network</code>.</li> <li><strong><code>NetworkAnalyzer</code></strong>: Moved to <code>bioneuralnet.network</code>; GPU-accelerated via PyTorch; added <code>hub_analysis</code>, <code>cross_omics_analysis</code>, <code>edge_weight_analysis</code>, <code>find_strongest_edges</code>, <code>degree_distribution</code>, <code>clustering_coefficient_gpu</code>, <code>connected_components</code>.</li> <li><strong><code>auto_pysmccnet</code></strong>: Phenotype-driven network construction via SmCCNet 2.0; supports CCA and PLS modes, now fully implemented in native Python, simplifying user experience and removing the R dependency.</li> </ul> <h3>Utils Module</h3> <ul> <li><strong><code>impute_omics</code> / <code>impute_omics_knn</code> renamed</strong>: Now <code>impute_simple</code> and <code>impute_knn</code>.</li> <li><strong><code>normalize_omics</code> renamed</strong>: Now <code>normalize</code>; supports <code>"standard"</code>, <code>"minmax"</code>, <code>"log2"</code>.</li> <li><strong><code>beta_to_m</code> renamed</strong>: Now <code>m_transform</code>.</li> <li><strong>New <code>feature_selection</code> submodule</strong>: <code>laplacian_score</code>, <code>mad_filter</code>, <code>pca_loadings</code>, <code>correlation_filter</code>, <code>importance_rf</code>, <code>variance_threshold</code>, <code>top_anova_f_features</code>.</li> <li><strong>New <code>data</code> functions</strong>: <code>data_stats</code>, <code>sparse_filter</code>, <code>nan_summary</code>, <code>zero_summary</code>.</li> <li><strong><code>clean_internal</code></strong>: New cleaning function with configurable NaN threshold.</li> </ul> <h3>DPMON Enhancements</h3> <ul> <li><strong><code>tune_trials</code></strong>: Already introduced in 1.2.2; now fully documented.</li> <li><strong><code>ae_architecture</code></strong>: New parameter; supports <code>"original"</code> and <code>"dynamic"</code> autoencoder architectures.</li> <li><strong><code>correlation_mode</code></strong>: New parameter; supports <code>"abs_pearson"</code> (default) and <code>"adaptive"</code> node feature computation.</li> <li><strong>Inner CV tuning</strong>: Ray Tune now performs epoch-synchronized inner k-fold cross-validation across all trials.</li> </ul> <h3>Datasets</h3> <ul> <li><strong>PAAD removed</strong> from built-in datasets.</li> <li><strong>Dataset size reduction</strong>: BRCA, LGG, and KIPAN datasets significantly reduced from ~4,000 omics features per dataset to 700 (400 methylation, 200 mRNA, 100 miRNA) using Laplacian Score filtering, replacing the previous ANOVA-F & Random Forest intersection strategy. This standardization was necessary to stay within the PyPI 100 MB package size limit (v1.2.2 reached 97.9 MB) and results in substantially faster installs and downloads for users.</li> </ul> <h3>Documentation</h3> <ul> <li><strong>Data Decision Framework</strong>: New comprehensive stage-by-stage parameter reference (<code>quick_start/data_framework.rst</code>).</li> <li><strong>Quick Start notebooks</strong>: New home for end-to-end <code>Quick_Start.ipynb</code> and <code>quick_start_bio.rst</code>.</li> <li><strong>Subgraph page</strong>: Updated case studies from KIPAN to TCGA-LGG and ROSMAP with full algorithm documentation.</li> <li><strong><code>network.rst</code></strong>: New dedicated page for the network module.</li> <li><strong><code>utils.rst</code></strong>, <strong><code>datasets.rst</code></strong>, <strong><code>index.rst</code></strong>, <strong><code>subgraph.rst</code></strong>: Major updates throughout.</li> <li><strong>README</strong>: GitHub readme updated to reflect all API changes, new images, and corrected function names.</li> </ul> <h3>Removed</h3> <ul> <li><code>gen_similarity_graph</code>, <code>gen_correlation_graph</code>, <code>gen_threshold_graph</code>, <code>gen_gaussian_knn_graph</code> from <code>bioneuralnet.utils</code>.</li> <li><code>graph_analysis</code>, <code>repair_graph_connectivity</code>, <code>find_optimal_graph</code> from <code>bioneuralnet.utils</code> (superseded by <code>NetworkAnalyzer</code> and <code>network_search</code>).</li> <li><code>impute_omics</code>, <code>impute_omics_knn</code>, <code>normalize_omics</code>, <code>beta_to_m</code> (renamed, see above).</li> </ul> <h3>Testing</h3> <ul> <li>Test suite updated to align with new <code>network</code> module and renamed utils functions.</li> </ul> <p><em>pasted from Changelog</em></p>
format Recurso digital
id zenodo_https___doi_org_10_5281_zenodo_19383738
institution Zenodo
language
publishDate 2026
publisher Zenodo
record_format zenodo
spellingShingle UCD-BDLab/BioNeuralNet: v1.3.0
Vicente Ramos
Mohamed Abdel-Hafiz
SundousHussein
<h2>[1.3.0] - 2026-04-01</h2> <h3>Network Module (<code>bioneuralnet.network</code>)</h3> <ul> <li><strong>New dedicated module</strong>: Network construction and analysis moved from <code>bioneuralnet.utils</code> to <code>bioneuralnet.network</code>.</li> <li><strong>Renamed construction functions</strong>: <code>gen_similarity_graph</code> -> <code>similarity_network</code>, <code>gen_correlation_graph</code> -> <code>correlation_network</code>, <code>gen_threshold_graph</code> -> <code>threshold_network</code>, <code>gen_gaussian_knn_graph</code> -> <code>gaussian_knn_network</code>.</li> <li><strong><code>NetworkAnalyzer</code></strong>: Moved to <code>bioneuralnet.network</code>; GPU-accelerated via PyTorch; added <code>hub_analysis</code>, <code>cross_omics_analysis</code>, <code>edge_weight_analysis</code>, <code>find_strongest_edges</code>, <code>degree_distribution</code>, <code>clustering_coefficient_gpu</code>, <code>connected_components</code>.</li> <li><strong><code>auto_pysmccnet</code></strong>: Phenotype-driven network construction via SmCCNet 2.0; supports CCA and PLS modes, now fully implemented in native Python, simplifying user experience and removing the R dependency.</li> </ul> <h3>Utils Module</h3> <ul> <li><strong><code>impute_omics</code> / <code>impute_omics_knn</code> renamed</strong>: Now <code>impute_simple</code> and <code>impute_knn</code>.</li> <li><strong><code>normalize_omics</code> renamed</strong>: Now <code>normalize</code>; supports <code>"standard"</code>, <code>"minmax"</code>, <code>"log2"</code>.</li> <li><strong><code>beta_to_m</code> renamed</strong>: Now <code>m_transform</code>.</li> <li><strong>New <code>feature_selection</code> submodule</strong>: <code>laplacian_score</code>, <code>mad_filter</code>, <code>pca_loadings</code>, <code>correlation_filter</code>, <code>importance_rf</code>, <code>variance_threshold</code>, <code>top_anova_f_features</code>.</li> <li><strong>New <code>data</code> functions</strong>: <code>data_stats</code>, <code>sparse_filter</code>, <code>nan_summary</code>, <code>zero_summary</code>.</li> <li><strong><code>clean_internal</code></strong>: New cleaning function with configurable NaN threshold.</li> </ul> <h3>DPMON Enhancements</h3> <ul> <li><strong><code>tune_trials</code></strong>: Already introduced in 1.2.2; now fully documented.</li> <li><strong><code>ae_architecture</code></strong>: New parameter; supports <code>"original"</code> and <code>"dynamic"</code> autoencoder architectures.</li> <li><strong><code>correlation_mode</code></strong>: New parameter; supports <code>"abs_pearson"</code> (default) and <code>"adaptive"</code> node feature computation.</li> <li><strong>Inner CV tuning</strong>: Ray Tune now performs epoch-synchronized inner k-fold cross-validation across all trials.</li> </ul> <h3>Datasets</h3> <ul> <li><strong>PAAD removed</strong> from built-in datasets.</li> <li><strong>Dataset size reduction</strong>: BRCA, LGG, and KIPAN datasets significantly reduced from ~4,000 omics features per dataset to 700 (400 methylation, 200 mRNA, 100 miRNA) using Laplacian Score filtering, replacing the previous ANOVA-F & Random Forest intersection strategy. This standardization was necessary to stay within the PyPI 100 MB package size limit (v1.2.2 reached 97.9 MB) and results in substantially faster installs and downloads for users.</li> </ul> <h3>Documentation</h3> <ul> <li><strong>Data Decision Framework</strong>: New comprehensive stage-by-stage parameter reference (<code>quick_start/data_framework.rst</code>).</li> <li><strong>Quick Start notebooks</strong>: New home for end-to-end <code>Quick_Start.ipynb</code> and <code>quick_start_bio.rst</code>.</li> <li><strong>Subgraph page</strong>: Updated case studies from KIPAN to TCGA-LGG and ROSMAP with full algorithm documentation.</li> <li><strong><code>network.rst</code></strong>: New dedicated page for the network module.</li> <li><strong><code>utils.rst</code></strong>, <strong><code>datasets.rst</code></strong>, <strong><code>index.rst</code></strong>, <strong><code>subgraph.rst</code></strong>: Major updates throughout.</li> <li><strong>README</strong>: GitHub readme updated to reflect all API changes, new images, and corrected function names.</li> </ul> <h3>Removed</h3> <ul> <li><code>gen_similarity_graph</code>, <code>gen_correlation_graph</code>, <code>gen_threshold_graph</code>, <code>gen_gaussian_knn_graph</code> from <code>bioneuralnet.utils</code>.</li> <li><code>graph_analysis</code>, <code>repair_graph_connectivity</code>, <code>find_optimal_graph</code> from <code>bioneuralnet.utils</code> (superseded by <code>NetworkAnalyzer</code> and <code>network_search</code>).</li> <li><code>impute_omics</code>, <code>impute_omics_knn</code>, <code>normalize_omics</code>, <code>beta_to_m</code> (renamed, see above).</li> </ul> <h3>Testing</h3> <ul> <li>Test suite updated to align with new <code>network</code> module and renamed utils functions.</li> </ul> <p><em>pasted from Changelog</em></p>
title UCD-BDLab/BioNeuralNet: v1.3.0
url https://doi.org/10.5281/zenodo.19383738